The demonstration
Building the session

Showing someone how, so they can see it

The hard part of demonstrating is that you cannot see the skill any more. This chapter breaks what your hands do into steps a stranger can follow, and asks of every step what must be visible and from where.

Start here Essentials

A demonstration built by somebody who can do the thing is built for somebody who can do the thing. This chapter takes it apart and rebuilds it for a person who cannot.

  1. Do the thing once at normal speed and write down every separate hand movement. Right after, while it is fresh (section 2).
  2. Mark the steps you did without thinking. Those are the ones you would have skipped (section 2).
  3. For each step, write what a person must see and from where. One line each (section 3).
  4. Find the step where your own hand blocks the view, and decide what to do about it. Turn, mirror, camera, or a second showing from the other side (section 3).
  5. Decide which steps you narrate and which you do in silence. Mark them on the sheet (section 4).

Minimum time: about 45 minutes, including doing the thing twice. Done when: the demonstration breakdown in your demonstration and practice sheets has one row per step, each naming what must be visible, and at least one step is marked as something you had not noticed you were doing.

1. You cannot see the skill any more Understand

The thing that makes you qualified to teach this is the same thing that makes the demonstration hard to build. You have done it enough times that most of it has stopped being a sequence of decisions and become one movement, and a movement you do not experience in parts is a movement you cannot describe in parts.

The University of Wisconsin-La Crosse's teaching guide on reducing cognitive load names the quality you need as cognitive empathy, and states the problem plainly: teachers are experts and students are novices, and the gap is hard for an expert to remember. Physical principle The same page describes the general constraint behind it: people can hold and manipulate only a few pieces of information at a time, and overload produces poor learning. Your demonstration is a stream of information arriving at the speed you can produce it, which is faster than a beginner can take it in.

Three failures follow, and they are the three this chapter fixes:

  • Merged steps. What you experience as "pick it up and set it" is four movements. People copy the two they saw.
  • Blocked views. Your hand goes where your hand goes, which is usually between the thing and the room.
  • Narration that runs ahead of the hands. You explain the next step while doing this one, because you know what is coming and they do not.

None of these is about confidence or presence. They are mechanical problems with mechanical fixes, which is good news if standing in front of people worries you: the work is on the bench, not on your personality.

2. Breaking it into steps

Here is the exercise, and it takes about twenty minutes. Do the thing at normal speed, exactly as you would alone, and then immediately write down every separate movement in order. Do not write while doing it; that changes the performance. Do it, then write.

You will write seven steps. Then do it again, slowly, and count: there will be twelve to fifteen. The gap between the two numbers is the part you teach badly, because the missing steps are the ones that have become automatic.

Three ways to find the missing ones:

  • Watch somebody who learned recently. Where their hands hesitate is a step you merged. Say nothing while they work, including encouraging noises, which are a form of correction.
  • Narrate to a wall. Do it while describing each movement out loud to nobody. The places where your description goes vague are the automatic steps.
  • Film your hands. A phone propped against a jar. Watch it at half speed with the sound off and count the movements.

Each step goes into a row of the demonstration breakdown sheet, with its own line for what you do, what you say, what must be visible and from where. The sheet has a column for the error this step causes, which you fill from memory: for each step, what goes wrong later if a person gets this one wrong. That column is what turns a list of movements into teaching.

Step as you first wrote itWhat it actually containsThe error if it is missed
Flour the benchTake a pinch, scatter from a height, spread with the flat hand, brush the excess to one sideToo much flour, dough slides instead of gripping, no tension in the shape
Pick up the doughScrape under with the bench scraper at a low angle, lift with both hands, let it hang for a secondDough tears, gas pushed out, shape fails at the seam
Shape itCup the hands, drag toward you, rotate a quarter turn, repeat four to six times, feel the skin tightenShape looks right and collapses standing, because there is no tension in the skin
Set it downSeam underneath, pinch the seam, place without dropping, leave spaceSeam opens in the oven

Four steps became sixteen movements and four named errors. This is the work of the chapter.

3. What must be visible, and from where

Fill this column before you fill the "what you say" column, because it is the one that changes where people stand and what you have to build.

For each step, answer two questions. What is the thing a person has to see to copy this? And from which position in the room can they see it?

The answers are often uncomfortable. The angle of a hand against a bench is visible from the side and invisible from above. The depth a blade goes is visible from the end and invisible from the front. A screen is visible to the first two rows. Test it The only reliable way to find out is to sit in each seat in the room and look, which is why Setting up a room people can actually work in asks you to sit in the worst one, and why the room plan is worth drawing before the day.

The sightline risk column on the sheet gives you five answers to choose from: clear from every seat, my hand is in the way, too small to see, over too fast, only visible from one side. Each has a standard remedy:

RiskRemedies, cheapest first
My hand is in the wayTurn the work rather than yourself; do that step with the other hand; narrate it and then show the result; do it twice, from two sides
Too small to seeBring people in close for that step only; do it at the back table in two halves; a phone on a stand projecting to a screen
Over too fastDo it twice: once at speed so they see what it looks like, once slowly so they see what it is
Only visible from one sideTwo showings, or move the group; mark the floor where to stand
Clear from every seatCheck it from the worst seat anyway, sitting down

Describing what you are doing out loud is not only a teaching move. The W3C's checklist for making events accessible instructs speakers to describe all relevant visual information, which means that a demonstration narrated well enough reaches somebody who cannot see it, and a demonstration performed in silence does not. Convention That is worth knowing before you decide which steps to do silently.

4. What to say, and when to say nothing

Narration is the second column of the sheet and it has two failure modes. Saying too much, so that the words compete with the thing they are about. And saying nothing, so that the room watches a performance and sees only its surface.

A workable default: narrate the steps that are invisible or counterintuitive, and stay quiet through the steps where watching is the whole job. The steps you marked as automatic are usually the ones to narrate, because they are the ones a person would not know to do.

Three rules that help:

  • Say the step as you do it, not before. Saying "now I am going to turn it" and then turning it splits attention. Turning it while saying "turning it a quarter" does not.
  • Name the thing to look at, then pause. "Watch the angle of my hand against the bench" followed by a silent two seconds is worth more than a sentence of explanation during the movement.
  • Keep the why for afterwards. Reasons are valuable and they compete with watching. Give the reason after the step, or at the end, and the room can attend to one thing at a time.
Limits

A demonstration shows what the thing looks like when it goes right. That is less than people assume. Watching a skill performed well does not transfer it, which is why practice and feedback are a separate chapter, and why a session that is all demonstration leaves a room that has seen rather than a room that can do. This chapter also cannot tell you how many times to show it: that depends on the skill, the room and your minutes, and no source in this manual's resource library gives a figure.

5. Peyton's four steps, and what the trial found Go deeper

The best-known structure for teaching a practical skill is Peyton's four-step approach. It is worth giving accurately, with the finding people leave out.

A 2016 randomized trial published in BMC Medical Education, Peyton's four-step approach for teaching complex spinal manipulation techniques, states the steps as follows:

  1. Demonstrate. The trainer demonstrates the skill at a normal pace, without additional comments.
  2. Talk the trainee through. The trainer demonstrates again while describing each substep in detail.
  3. Trainee talks trainer through. The trainer performs the skill a third time, following the trainee's description.
  4. Trainee does. The trainee performs the skill alone.

Now the results from that same trial, which randomized 95 third- to fifth-year medical students. There was no difference between the groups on the multiple-choice exam. The Peyton group scored significantly higher on the practical exam. And a second practical exam showed a significant decline in competency regardless of the type of instruction: in the paper's own words, skills retention was equally low for both techniques.

A 2025 randomized simulation study of the approach in structured advanced cardiac life support training, published in the Journal of Medical Education and Curricular Development, concludes that integrating the approach enhances procedural skills, adherence to the algorithm and retention in advanced stages of medical education. That paper also states its own counterweight: a few studies indicate the approach is not more effective than other approaches, and, in its words, there is limited evidence supporting the superiority of Peyton's approach over other teaching methods.

Convention So here is the honest position this manual takes. The four steps are a good default, for two reasons that have nothing to do with the evidence for the approach as a whole. Step two forces you to narrate every substep, which is exactly the work this chapter asks for. Step three hands the describing to the learner before they touch anything, which surfaces what they have misunderstood while it is cheap to fix. Use them because of what they make you do, rather than because they are proven to make learning stick, and plan for the retention problem separately: what people do in the week afterwards matters, which is why the job aid in Slides, handouts and materials: what you give them and when is not an afterthought.

Step three is the one people drop, and it is the cheapest to keep. With a group, one person talks you through while everyone watches, and the mistakes in their description are the mistakes in the room's understanding.

6. Cognitive load and worked examples Go deeper

The UW-La Crosse guide describes cognitive load as the mental effort a task takes, in three sources. Intrinsic load is the task itself: how many things have to be held at once to do it. Extraneous load comes from the instructional context, for example disorganized instruction: it is load you added. Germane load is described on that page as deep learning and sense-making. Convention Some researchers dispute germane load as a separate category, so this manual uses the first two, names the third as disputed, and builds nothing on it.

What follows for a demonstration is concrete. You cannot reduce intrinsic load, because the task is the task, though you can break it into parts and practice them separately. You can reduce extraneous load, and most demonstrations carry a lot of it: things on the bench that are not in use, a story about last time, the names of three alternative tools, a slide still showing behind you.

The same page lists five moves: build on prior knowledge, segment the material, increase coherence and cut irrelevant material, automate routines, and scaffold complex tasks. It defines a worked example as a form of scaffolding in which the whole solution is available. Your demonstration is a worked example, and the question that follows is when to take it away: a person who can only perform with the model in front of them has not got it yet, so the second attempt should happen without the example in view.

7. Mayer's principles, and what they were tested on Go deeper

If your demonstration involves a screen, you will meet Mayer's multimedia principles. The University of Washington School of Medicine's page on principles for teaching with multimedia groups them under three goals: reduce extraneous processing, manage essential processing, foster generative processing. The ones that bear on a demonstration are coherence (exclude extraneous material), signaling (highlight the main ideas), redundancy (narration with the animation rather than narration plus identical on-screen text), spatial and temporal contiguity (put related words and pictures near each other, and together in time), and segmenting (let the learner control the pace of the parts).

Physical principle The caveat matters as much as the list. These were tested on narrated animations and multimedia lessons, not on a person standing at a bench with a group around them. Temporal contiguity, saying the words while the thing happens, transfers to a live demonstration fairly obviously. Segmenting, in the sense of learner-paced parts, does not transfer to a live demonstration, because the pace is yours. Treat the list as a source of good questions about a live demonstration rather than as findings about one. Slides, handouts and materials: what you give them and when uses the same source for slides, where it does apply, and sets the redundancy principle against a finding that pulls the other way.

8. Cameras, mirrors, second demonstrations and latecomers Go deeper

Four practical problems, with what actually works.

Twelve people and one pair of hands

Past about eight people, a bench demonstration reaches the front. The cheap fixes, in order of cost: bring everybody in close and stand them in a horseshoe for that step only; do the demonstration twice, at two tables, with half the room each; prop a phone on a small stand over the work and mirror it to a screen. A mirror angled over the bench works and is awkward to position; it is worth testing before the day rather than on it.

The second demonstration at the back table

If you have a helper who can do the thing, the most effective arrangement is two demonstrations happening at once at two tables, six people each. It halves the group size for the step that most needs small groups. The cost is that your helper has to be able to do it, and the two demonstrations will differ in small ways, so agree the steps with them beforehand from your own breakdown sheet.

Repeating for latecomers

Somebody arrives fifteen minutes in. Repeating the demonstration costs the room ten minutes and teaches the latecomer less than it looks, because they missed the setup. Two better moves: pair them with somebody who was there and ask that person to talk them through it, which is step three of Peyton's sequence and is good for both; or give them the job aid and a two-sentence catch-up, and let them join the practice late.

Demonstrating the wrong way

Showing the common mistake is tempting and it has a known risk: people remember the picture and lose the label. If you do it, show the right way first, show the wrong way briefly, name it as wrong while you do it, and finish on the right way. For a safety-relevant error, consider describing it rather than performing it.

9. See it See it

Fictional example

Priya takes her shaping apart. Priya Raman has two hours, ten people and five benches at the community kitchen. She has shaped loaves for six years and has not taught before.

What she wrote from memory: flour the bench, pick up the dough, shape it, set it down. Four steps.

What filming her hands showed: sixteen movements, and three she had no memory of making. She lets the dough hang for a second after lifting it, which is how she judges whether it is slack. She wipes her palm on her apron before the final turn, because a damp palm drags. And she taps the bench twice before setting the loaf down, which turns out to be how she checks the flour is even.

Known: ten people, five benches, a bench she can stand at with people on three sides.

The blocked step: the cupping movement is visible from the side and hidden from the front, and the front is where six of the ten will be standing. Her decision: do that step twice, turning ninety degrees between, and say so out loud.

What she narrates: the hang, the wipe, the tap, the angle of the hand. All four are things she did without noticing.

What she does in silence: the four shaping drags, after saying "watch the skin on top get tighter." The silence is deliberate and about three seconds longer than feels comfortable.

Assumed, and whose assumption it is: that ten people can see from a horseshoe at one bench. Priya's assumption, and she tests it by asking Ann to stand at the back while she runs through it. On these assumptions the single demonstration holds, and if it does not, the fallback is two showings at two benches.

Open, and who settles it: whether the kitchen's benches can be moved into a horseshoe. The kitchen manager settles that, and Priya asks before she plans the room.

10. Practice Practice

Minimum: the thing done once, every movement written, and the "what must be visible" column filled. Time: about 45 minutes. Materials: whatever you teach with, a phone to film with if you have one, and the demonstration breakdown sheet in your demonstration and practice sheets.

  1. Do it at normal speed, then write the steps. Do not write while doing. Count what you wrote.
  2. Do it again slowly, or film it, and count the movements. Add the missing rows. Mark the ones you had not noticed.
  3. Fill the "what must be visible" and "from where" columns. One line per row, before you write any narration.
  4. Mark the sightline risk on each row, and pick a remedy for every row that is not clear from every seat.
  5. Fill the error column. For each step, what goes wrong later if a person gets this one wrong. This is the column you will use most in the room.
Go deeper

Run the four-step sequence with one person before the session: demonstrate in silence, demonstrate while describing, have them describe while you do it, then have them do it. The third step is where you learn the most, because their description is a transcript of what your demonstration communicated. Write the gaps straight into the sheet. This takes about twenty minutes with a willing friend and it is the single most useful rehearsal in this manual.

11. Your session Your session

12. Check your work Check your work

Criteria

  • Does the sheet have more steps than you first wrote from memory?
  • Is at least one step marked as something you had not noticed you were doing?
  • Does every row say what must be visible and from where?
  • Does every row with a sightline risk have a remedy written next to it?
  • Could somebody read your rows and know which parts you will narrate and which you will do in silence?

Worked example

Priya's sheet went from four steps to sixteen, three of them marked as unnoticed. Every row names what must be visible, and the two rows with a blocked view have the remedy written: turn ninety degrees, show twice. Her narration marks are on four rows. On the fifth criterion her sheet is thin: she has marked what she will say but not where the silences are, and the silences are the part she will fill with talking if she has not planned them.

Common mistakes

  • Writing the steps from memory and stopping. The memory version is the merged version.
  • Filling the narration column first. It is the comfortable column, and it hides the sightline problem.
  • One demonstration for a group that cannot see it. Count who can see before deciding.
  • Explaining the reason during the movement. Reasons compete with watching. Put them after.
  • Treating the demonstration as the teaching. It is the setup for the practice, and the practice is where people get better.

What is still unresolved

This chapter does not tell you how many attempts people need afterwards, or what one attempt is: that is Designing the part where they do it. It does not tell you whether they got it, which is Checks for understanding: finding out whether it landed. Where people stand and what they can see from there is settled in Setting up a room people can actually work in and drawn in the room plan.

13. Knowledge check and scenario

Why write the steps after doing the thing rather than while doing it?
Writing while performing changes the performance: you slow down, you consciously sequence, and the automatic steps become deliberate ones, which hides exactly what you are trying to find. Doing it at normal speed and writing immediately afterwards gives you your real version, and the gap between that list and a filmed count is the part of the skill that has become invisible to you.
What did the 2016 trial of Peyton's four steps actually find?
Three things. No difference between groups on the multiple-choice exam. A significantly higher score for the Peyton group on the practical exam. And, on a second practical exam, a significant decline in competency regardless of instruction type, which the paper describes as skills retention being equally low for both techniques. A 2025 trial that favors the approach states in its own words that there is limited evidence for its superiority over other methods.
Why fill the "what must be visible" column before the narration column?
Because it is the column that changes the physical arrangement of the session: where people stand, whether you need a second showing, whether a phone and a screen are required. Narration is easy to write and easy to change on the day. Sightlines are neither, and a beautifully narrated step that half the room cannot see has taught half the room nothing.
(Go deeper) What is extraneous load, and where does a demonstration usually carry it?
It is the part of the mental effort that comes from the instructional context rather than from the task, which means it is load you added. In a demonstration it is usually: tools on the bench that are not in use, an anecdote running under the movement, alternative methods named in passing, a slide still showing behind you, and reasons given during the step rather than after it. Cutting it costs nothing and is the cheapest improvement available.
Twelve people, one bench, and the key movement is only visible from the side. You have no screen and no helper. What do you do?
This is what usually happens and it is a decision made by default. Six of the twelve will watch a shoulder. If you do take this route, take it deliberately: do the key movement twice, turning between, and say which side to stand on for each.
Halving the group for the step that needs small groups is the strongest move available without equipment. It costs you a repeat of six or eight minutes, and it needs the other six to have something specific to do, which is the same question the practice chapter asks about people waiting for a station. Write that task down rather than inventing it in the room.
A recording solves the sightline problem and gives up the thing a live demonstration is for: people can stop you, you can respond to a face, and you see what they are looking at. A good compromise is live for the whole, recorded close-up for the one movement nobody can see, played twice.

Checklist

Records your learning only; it is not a qualification.